Budget Fitness Bands
We shortlisted affordable, well-specced bands first. The problem wasn't hardware — it was software. Most run closed, vendor-locked operating systems with no real path to installing our own app.
Back to the blog
Engineering
Why We Chose the Galaxy Watch (For Now)
Building custom hardware is slow, and it's the wrong place to spend our first year. Here's how we picked the watch we're building on instead.
A custom wearable sounds appealing — until you're a year into tooling and certification before a single detection model has seen real-world data. We wanted the opposite: a device we could put on a wrist this quarter, so the software could start learning immediately. That meant evaluating what's already on the market.
The Wear OS Requirement
To run our own detection logic on-device, we needed a real app platform with raw sensor access — not a companion app locked to the manufacturer's ecosystem. That narrowed the field to Wear OS.
Two Finalists
We came down to two Wear OS watches with the sensor set and specs to support real-time detection: the OnePlus Watch 2R and the Samsung Galaxy Watch.
The Galaxy Watch 7
We finalised on the Samsung Galaxy Watch 7 — a mature sensor suite, solid Wear OS tooling, and documentation we could actually build against.
It's a trade-off we made with eyes open: no mainstream Wear OS watch, including this one, has a true electrodermal activity sensor. Here's why that matters, and how we're working around it →
What an Off-the-Shelf Watch Buys Us Today
A real display, out of the box
Full control over on-device app logic
Raw access to motion & heart-rate sensors
Months, not years, to first pilot
Custom hardware — potentially with clinical-grade sensors like true EDA — is still on the roadmap. For now, this gets a working system on real wrists sooner, which is what actually improves the detection models.
See the hardware in motion.
From the wrist to the cloud, here's the full detection pipeline.
How Your Watch Talks to the Cloud